Skip to main content

​
PSTN Voice Go-Live Checklist

The Contact-center (voice/CCaaS) go-live runbook covers the contact-center layer — softphone, ACD queues, IVR, wrap-up, and the cutover plan. This page covers the PSTN layer underneath it: the numbers, the sender identity, the routing, the recording, the compliance, and the voicemail that carry live voice onto the public network. Run this in addition to the CCaaS runbook when your first live callers reach a real DID over a real trunk. Both pages stack on the platform-wide Production go-live checklist; finish the platform gates first.

​
1. Scope

This checklist applies when raw voice (inbound DID, outbound PSTN dial, or a SIP trunk to your own PBX) moves off sandbox numbers onto live traffic. Cover the ten checks in order; skip a check only when its surface is genuinely unused (for example, no SIP trunk if you route API-direct). What goes live:
  • Numbers — toll-free (800/888/877/866/855/844/833/833) and local DID, in E.164, provisioned through the numbers console and wired to a route.
  • Sender identity — the number presented on outbound: E.164, an alphabetic Sender ID where the market allows it, a short code, or a branded caller-ID (CNAM / RCD logo) on supported carriers.
  • Routing — how a call reaches Orbit: a SIP trunk (BYOC, BYOB), the API direct path, or a host-associated DID that rings a registered softphone or PBX extension.
  • Compliance — STIR/SHAKEN attestation posture, E911 emergency address on file, and the ATP/bundle tenancy where your jurisdiction requires it.
  • Recording and retention — call recording on or off, the retention window, and where the media lives (Orbit-managed or BYO storage).
  • Voicemail — the mailbox that catches unanswered, declined, and overflow callers, and who owns it.
  • Queue coverage — which inbound DIDs terminate into an ACD queue and which ring a direct line.

​
2. Preflight — the ten checks

Run these in order; each proves the previous one. Put a named owner on every line before launch day.
  1. Numbers — toll-free + DID reality-check. Confirm each live DID exists, is active, and is in E.164 (+1…). Buy or port per Buy and provision numbers and Number porting; verify the port landed before you point traffic at it. Owner: Operations.
  2. Sender identity — E.164 + alphabetic + short codes + branded caller-ID. Pick the outbound presentation per Sender-ID pre-registration matrix: E.164 where only numeric is allowed, an alphabetic sender where the market permits (and pre-register it where the market forces it), a short code for high-volume programmatic, or a branded caller-ID. Owner: Compliance/Operations.
  3. Caller-ID branding — CNAM and RCD. Register the branded name (CNAM, 1–15 chars) per CNAM and caller ID branding, and configure the RCD logo per Set up branded calling (RCD) so the logo renders on the handset of a carrier that supports Rich Call Data. Owner: Compliance/Operations.
  4. Routing — SIP trunk / API direct / host-associated. If you trunk, connect per Connect a SIP trunk (BYOC) and validate before save per SIP trunk first call; if you dial API-direct, the softswitch handles the trunk; if you ring a registered PBX/softphone, the DID is host-associated. Confirm the path the call will actually take. Owner: Engineering.
  5. Compliance — STIR/SHAKEN attestation. Resolve to full A-attestation where your numbers qualify per Assemble your STIR/SHAKEN attestation posture; confirm with the STIR/SHAKEN Attestation Checklist that the floor is set, the delegate-certificate lifecycle is owned, and the per-DID inbound floor is on. Owner: Compliance.
  6. Compliance — E911 emergency address. Keep an emergency address on file where the jurisdiction requires it per E911 emergency drill mode; run the 933 drill to confirm the address reaches the right PSAP-adjacent destination. Remember: Orbit blocks the live short emergency codes (911/112/999/000) at dispatch — they never reach a carrier; agents dial emergencies from a regular phone. Owner: Compliance.
  7. PBX / desk-phone registration. Register each hardware desk phone by MAC per Desk phones: zero-touch provisioning, or pair each PBX extension with a SIP credential; confirm the registration stands (200 OK REGISTER) before launch. Owner: Engineering.
  8. Teams and ring-groups. Staff the team per Teams and wire the ring-group per Ring groups so a DID that should hunt across a set of destinations does so with the strategy you picked. Owner: Operations.
  9. Voicemail boxes. Stand up the overflow mailbox per Set up voicemail boxes: attach a greeting (upload or TTS), wire the DID or queue overflow to the box, and confirm full-read fencing and MWI. Decide and record who owns the box — the agent, the team, or a shared department inbox. Owner: Operations.
  10. Recording and retention. Decide on/off per Recording lifecycle operations, set the retention window, and choose Orbit-managed or BYO storage; confirm the consent prompt or beep tone fires where the jurisdiction requires it. Owner: Compliance/Operations.

​
3. Per check — smoke validation

What proves each check passes before you cut traffic over:
  • 1 Numbers: phone the DID from an outside line; the call resolves to the route you attached (IVR, queue, softphone, or voicemail), not a carrier intercept.
  • 2 Sender identity: place one outbound call to a test handset; the displayed number matches the E.164 you configured, the alphabetic sender renders where allowed, or the short code shows where provisioned.
  • 3 Caller-ID branding: the CNAM name renders on a carrier that supports it; the RCD logo renders on a supported handset — both verified on a live PSTN call, not a softphone-to-softphone loop.
  • 4 Routing: the inbound DID resolves through the SIP trunk (200 OK INVITE on the trunk), the API-direct path completes, or the host-associated DID rings the registered device — whichever you picked.
  • 5 STIR/SHAKEN: the attestation header on the wire reads full A (not B or C partial); the Checklist tool reports the same level the carrier sees.
  • 6 E911: the 933 drill reaches the test destination and the address-of-record it returns matches what you filed; the drill does not touch a live PSAP.
  • 7 PBX / desk-phone: each desk phone shows registered; a 200 OK REGISTER appears in the SIP trace; an inbound call to the extension rings the phone.
  • 8 Teams / ring-groups: a call to the team DID hunts the group in the strategy you set (ring-all, round-robin, longest-idle) and lands on an available member.
  • 9 Voicemail: a call that overflows or is declined hits the box, the greeting plays, a message is left, and MWI lights the phone — and the box owner you assigned is the one who retrieves it.
  • 10 Recording / retention: a recorded call produces a media file, the retention date is set, and the consent prompt or beep fires where required; a non-recorded call produces nothing.

​
4. Failure modes

The four launch-day failures that bite when you skip a check:
  • Caller-ID blacklisting. The outbound number carries a spam/scam label from a prior owner; recipients decline or carriers filter it. The fallout: live answer rates drop and the dial looks like junk. Remediate per CNAM and caller ID branding — dip the carrier CNAM database, confirm the label, and request a clean re-registration; for persistent labels, retire the number and provision a fresh one. The prevention is check 2 and 3 before the first dial.
  • PBX not registering. The desk phone or PBX sits unregistered; inbound DIDs to it fail with a 404/480, or the trunk shows no registration. Triage with a minimal PCAP — sip.set the capture to the trunk’s SIP signalling IP and look for the REGISTER/401/REGISTER-200OK triplet; a missing 200 OK points at auth, a 401 loop points at the credential, and no traffic at all points at the IP-allowlist or the transport (TLS vs ws). Follow SIP trunks troubleshooting.
  • STIR attested as partial. The header reads B or C instead of A; downstream carriers flag the call. The cause is almost always an unresolved ownership level or a missing delegate certificate on a BYON number. Re-run Assemble your STIR/SHAKEN attestation posture, confirm the floor, re-enroll the delegate certificate, and re-read the header on the wire before you trust the report.
  • E911 lookup refusal. The emergency address is missing or stale and the jurisdiction requires it; the call is refused or routed to the wrong destination. Re-file the address per E911 emergency drill, re-run the 933 drill, and confirm the returned address before you let a live DID carry traffic.

​
5. Sign-off record

Record who approved each check before launch and where the evidence lives. Put a named owner on every line; an unnamed line is an unowned risk.
  • Operations signs off checks 1, 2, 8, 9 (numbers, sender identity, teams/ring-groups, voicemail) and the queue-coverage decision.
  • Compliance signs off checks 3, 5, 6, 10 (caller-ID branding, STIR/SHAKEN, E911, recording/retention) and the per-country tenant policy.
  • Engineering signs off check 4 and 7 (routing, PBX/desk-phone registration) and the cutover/rollback path.
File the evidence in your compliance binder: the STIR/SHAKEN posture snapshot, the E911 drill acceptance, the recording-consent configuration, the caller-ID verification, and the per-DID routing record. The binder is the proof you ran the ten checks — not the launch day itself. A launch that skips the binder is a launch that skips the checks.